OpenAI makes ChatGPT Health available to all U.S. users

OpenAI is done testing. Starting this week, ChatGPT Health — the feature built for medical questions — is rolling out to every logged-in US user over 18, across the free, Go, Plus, and Pro tiers, on web and iOS. The company had been piloting this through a dedicated hub since January, letting people plug in data from Apple Health, Function, and MyFitnessPal. Back then, OpenAI said 230 million health-related queries were hitting the chatbot each week. That figure is now 300 million.

The expansion arrives one day after a Florida pastor filed suit against OpenAI, alleging the chatbot told him not to see a doctor and that the advice nearly cost him his life. OpenAI has pointed to its own terms of service in response, which state plainly that the product isn't meant for diagnosing or treating any health condition. The company also told the New York Times it's working on making its medical and health answers safer going forward.

What's actually new here isn't just the wider rollout — it's where the health smarts now live. Previously, you had to go into the dedicated hub to get health-aware answers. OpenAI found that during testing, 70% of health-related questions were being asked outside that hub entirely, in ordinary chat. So now the system can pull from your connected health data — including records synced from Epic, Oracle Health, One Medical, and Function Health — no matter where in ChatGPT you ask. Ask about a recipe and an allergy, and it can factor in what it already knows about you.

On the model side, OpenAI says its newest and smallest model, GPT 5.6-Luna, beats the previous GPT 5.5 on HealthBench, the company's own open-source benchmark for grading how well language models handle medical questions. OpenAI also says it doesn't train on user data and works with physicians to refine these answers. Still, the company is telling users to double-check what they're told and lean on actual medical professionals for real decisions — a line that sits awkwardly next to a product now baked into every conversation.

OpenAI isn't alone in this bet. Despite research showing AI chatbots remain unreliable for medical advice, both Anthropic and Google have pushed ahead with their own health-focused AI features. The rush suggests the industry has decided the demand — hundreds of millions of health queries a week, apparently — is too big to leave on the sidelines, liability questions notwithstanding.